Download Arbitrage Pricing Theory - Lahore School of Economics

The Lahore Journal of Economics
10 : 1 (Summer 2005) pp. 123-139
Arbitrage Pricing Theory:
Evidence From An Emerging Stock Market
*
Javed Iqbal and Aziz Haider
**
The development of financial equilibrium asset pricing models has
been the most important area of research in modern financial theory. These
models are extensively tested for developed markets. This paper examines the
validity of the Arbitrage Pricing Theory (APT) model on returns from 24
actively trading stocks in Karachi Stock Exchange using monthly data from
January 1997 to December 2003. Explanatory factor analysis approach
indicates two factors governing stock return. Pre-specified macro economic
approach identifies these two factors as the anticipated and unanticipated
inflation and market index and dividend yield. Some evidence of instability
is found. The overall finding of two significant priced factors at least for a
sub period supports APT for an emerging capital market.
1. Introduction
The applications of financial equilibrium models have been very
intensively investigated. These applications are used for various purposes
such as predicting common stock systematic risk and defining the cost of
capital. The traditional equilibrium model, the capital asset pricing model
(CAPM) of Sharp (1964), Linter (1965) and Mossin (1966) assume that
stock returns are generated by a one-factor model, where the factor
represents the market portfolio of all risky assets. Empirical tests of the
CAPM have produced mixed results. The critical point in the estimation of
the CAPM is the difficulty of measuring the true market portfolio. Due to
the severe problems in the testing the CAPM (Copeland and Weston, 1988)
a number of the other models have been proposed.
Arbitrage pricing theory, developed by Ross (1976) proposes that
there are several sources of risk in the economy that cannot be eliminated
by diversification. These sources of risk can be thought of as related to
*
PhD student at Monash University, Department of Econometrics & Business Statistics
Australia. E-mail: [email protected]
** MSc student (Final) Department of Statistics, University of Karachi. E-mail:
[email protected]
124
Javaid Iqbal and Aziz Haider
economy wide factors such as inflation and changes in aggregate output.
Instead of calculating a single beta, like the CAPM, arbitrage pricing theory
calculates many betas by estimating the sensitivity of an asset’s return to
changes in each factor.
The arbitrage pricing theory assumes that a security return is a
linear function, not only of one, but also a set of common factors. The APT
thus indicates that the risk premium for an asset is related to the risk
premium for each factor and that as the asset’s sensitivity to each factor
increases, its risk premium will increase as well. The APT predicted that the
prices of all risky assets in the economy conformed to the condition of no
arbitrage. No arbitrage mean that an individual holding a well-diversified
portfolio could not earn any additional return merely by changing the
weights of the assets included in the portfolio, holding both systematic and
unsystematic risk constant. The APT states that there is a set of underlying
sources that influence all stocks returns. The stock return is a linear
function of a certain number; say k, of economic factors, while these factors
are unobservable and not meaningful.
According to Chen et al. (1986), these risk factors arise from
changes in some fundamental economic and financial variables such as
interest rates, inflation, real business activity, a market index, investor
confidence etc.
The APT thus starts with the assumption that returns on any stocks,
Rit , are generated by a k-factors model of the following for
R it = E (R i ) + bi1 F1 + b i 2 F 2 + ... + bik F k + ε i
---
(1)
Where E ( R i ), i=1,2,3…n, is the expected return of the stock i. Fj
(j=1,2,3…k,) are unobserved economic factors. bij is the sensitivity of the
security i to the economic factors j and εi are the unique risks of the stocks
(uncontrolled factor) i-e a random error term with mean equal to zero and
variance equal to σ ei2 .
Ross (1976) showed that if the number of stocks is sufficiently
large, the following linear risk-return relationship holds.
E (Ri ) = λ o + λ1bi1 + λ 2 bi 2 + ... + λ k bik
…
(2)
Where λ o is a constant risk less rate of return (the common return on all
zero-beta stocks), and λ j , j = 1,2,..., k represents, in equilibrium, the risk
Evidence from an Emerging Stock Market
125
premium for the jth factor. The mean zero random common factors can be
thought of as representing unanticipated changes in fundamental economywide variables. The sensitivity coefficients measure the magnitude and
direction of the reaction in asset returns.
In order to test the APT empirically, there are two main
approaches. First, one can simultaneously estimate the asset sensitivities and
unknown factors by exploratory factor analysis on stock returns. In that case
a theory does not predict the exact content or even the number of relevant
factors. Alternatively, we could try to specify prior general factors that
explain pricing in the stock market. Such macroeconomic variables could be
those affecting either future cash flows on companies or future risk-adjusted
discount rates. It is generally accepted that the trend of pre-specifying
factors seems to be a promising avenue of research in the search for
meaningful factor structure.
The factor analysis-based empirical tests of the APT on US data
have produced relatively mixed results. In their seminal paper, Roll and Ross
(1980) tested the APT for the period 1962-72. They used daily data for
individual equities listed on the New York Stock Exchange. They concluded
that at least three and probably four priced factors were found in the return
generating process.
Chen (1983) discovered that the APT seems to outperform the
traditional CAPM when evaluated by explanatory power on stock returns. He
investigated stocks using daily return data during the 1963-1978 period
from the New York Stock Exchange. He compared the empirical
performance of the APT with that of the CAPM.
More studies have found a number of critical issues when testing
the theory. For example, it has been found that the number of factors seems
to increase when the number of investigated securities increases. There is a
paucity of research evaluating the validity of the APT in non-US stock
markets. The sparse European results of the APT include these reported in
Diacgiannis (1986), Abeysekera and Mahajan (1988), Rubio (1988),
Ostermark (1989), Yli-Olli and Virtanen (1989), and Yli-Olli et al. (1990).
Concerning the Scandinavian results, Ostermark (1989) reported APTdominance on Finnish as well as Swedish data. Yli-Olli et al. (1990) found
three stable common factors across these two neighbouring countries, for
the period 1977-1986, using monthly data. They used the principal
component analysis to get the factor loadings, then cross-sectional OLS
regressions were applied for the three factor solutions to test how many
factors were priced in the two countries.
126
Javaid Iqbal and Aziz Haider
An alternative to the traditional approach is to specify a priori, on the
basis of the theory, the general factors that explain pricing in the stock
market. In this case the common factors are first measured using pre-specified
macroeconomic variables, and asset sensitivities to these factors are estimated
using time series regressions. In their seminal paper, Chen et al. (1986) found
that the spread between long-term and short-term interest rates, expected and
unexpected inflation, industrial production and spread between high and lowgrade bonds are priced in the generating process of stock returns in the US
stock market. These state variables have also been found to be important in a
number of other studies on US data such as Chen (1989).
Martikainen et al. (1991) tested APT for the Finish Stock Market using
monthly data. They used two different approaches: an exploratory factor
analysis and a pre-specified macroeconomic factor approach. They tested how
many factors there were that affected finish stocks in the two time periods
1977-81 and 1982-86. In the first step of the test they used principal
components analysis and varimax rotation to get the factor loadings. Then,
OLS regressions were made where factor loadings were independent variables
and the average return on stock was the dependent variable. The purpose was
to find how many factors that were priced in the market. In the second step
of the test they used 11 pre-specified macroeconomic factors to test the APT
model. They used different stock market indices, price indices, interest rates
and other national economic variables such as the GNP and money supply.
They could find only one priced factor for the first subperiod. In the second
subperiod all of the factors become priced. This was an encouraging result
that supported the theory that the equilibrium stock returns were generated
by an economic factor model.
Loflund (1992) found that international factors such as unanticipated
changes in real exchange rates, inflation and unanticipated changes in future
foreign economic activity or export demand should be important. National
factors such as unexpected inflation, unanticipated changes in the shortterm interest rate, the term structure of interest rates and unexpected
changes in domestic real production should be important.
Booth et al. (1993) tested the APT for US, Finnish and Swedish stock
returns during the 1977-86 period, using monthly data. They tested the intracountry stability of the factor patterns over time and across different samples.
They investigated the cross-sectional similarities of the factor patterns of
twelve 30-stock samples. They used transformation analysis to test the
stability. The empirical evidence indicated that two stable common factors in
different samples could be found. An interesting observation was that the
factors were very often produced in different order in different samples.
Another important finding was that there existed two common factors across
Evidence from an Emerging Stock Market
127
the first US sample and Finnish and Swedish samples. Thus, the two common
factors obtained have been international by nature. The results implied that
for Finland the APT performed relatively poorly and for US and Swedish data
one to two priced factors were identified.
For developing capital markets in general and Pakistani markets in
particular empirical evidence on equilibrium models are few. Khilji (1993) and
Hussain and Uppal (1998) investigated the distributional characteristics of stock
return in the Karachi Stock Exchange concluding that the return behavior
cannot be adequately modeled by a normal distribution. Hussain (2000) found
no evidence of the day of the week anomaly and concluded that for the period
January 1989 to December 1993 the absence of this predictability pattern
implied efficiency of the market. Ahmad and Zaman (2000) using sectoral
monthly data from July 1992 to March 1997 found that some of the CAPM
implications are valid in the Karachi Stock Market. They found evidence in
favor of positive expected return for investors but speculative bubbles were also
indicated. Khilji (1994) found that the majority of return series are
characterized by non-linear dependence. Ahmad and Rosser (1995) used an
ARCH-in-Mean specification to study risk return relationship using sectoral
indices. Attaullah (2001) tested APT in the Karachi Stock Exchange using 70
randomly selected stocks employing monthly data from April 1993 to
December 1998. Out of 11 macroeconomic factors he found unexpected
inflation, exchange rate, trade balance and world oil prices were sources of
systematic risk. He used Iterative Non Linear Seemingly Unrelated Regressions
technique. The present study provides another more recent evidence from
monthly data from January 1997 to December 2003. With a relatively greater
sample this study employs two different factor analysis techniques and stability
analysis is also performed. Moreover macroeconomic variables used are also
greater in number and regional market indices are also included.
The remainder of this paper is organized as follows. Section II, the
data used in this research effort is introduced. The empirical part of the study
is divided into two sections. Section III includes testing the APT using
traditional exploratory factor analysis approach. In Section IV macroeconomic
factors are identified using 16 pre-specified macroeconomic variables by
reducing the dimensionality of these variables using factor analysis. The APT is
also tested using these macroeconomic factors. Section V concludes.
II. The Data
The data consist of 24 actively traded stocks from the Karachi Stock
Exchange and the general market index (KSE-100), covering the period from
January 1997 to December 2003. Data on individual stocks regarding
128
Javaid Iqbal and Aziz Haider
closing prices was obtained from the Karachi Stock Exchange. These 24
stocks are the most active stocks with approximately 80% weight of
aggregate market capitalization of KSE 100 index companies. We have
collected the monthly data. In order to analyze the stability of the factors in
the APT, the period is divided into two subperiods
The first subperiod is from January 1997 to December 1999; the
second is from January 2000 to December 2003. One reason for breaking the
sample is stability testing of our results. Moreover the second period is more
volatile. In this period KSE attained its highest level of index value and market
capitalization. It is claimed to be the best performing capital market in the
world. Therefore we need a large sample for reliable estimates, while the first
period January 97 to December 99 is relatively smooth. The break up of the
sample can also be seen as pre-Musharaf government and the current
government. President Musharaf Government’s intended or unintended
economic, financial and foreign policies due to the 9/11 event have brought
drastic changes in the economic horizon. So it will also be tested whether
stock return behavior has changed in the two subperiods.
The returns have been measured using the first difference of monthly
logarithmic price indices. There are 16 macroeconomic variables, including
inflation measured by Consumer Price Index and Wholesale Price Index, a
measure of real economic activity. Ideally GDP should be used for this purpose
but the monthly data on GDP are not yet available for Pakistan therefore
manufacturing production index has been used to capture real economic
activity. Interest rate measured by 90 day T-bill of SBP, Money market rate, a
long-term interest rate yield on 10-year Pakistan Investment Bond are also
investigated in the analysis. When selecting the macroeconomic variables, they
have been chosen using the criterion that they should affect the rate of return
or future cash flow expectations of the firm share. All the variables are studied
using the first differences of the logarithmic forms of indices.
Inflation has been proxied by two indices measuring the wholesale
prices and the consumer prices. These variable are included in the study
since the classical Fisherian theory implies that the common stocks should serve
as an effective inflation hedge during expected inflation (see Mishkin, 1997). It
is generally observed that stock returns are negatively related to expected
inflation, unexpected inflation and change in expected inflation in several
countries (Asprem, 1989; Wasserfallen, 1989). When expected inflation rises,
interest rates will rise. Fama (1981), Geska and Roll (1983), Ram and Spencer
(1983), Stulz (1986), and Kaul (1987) all attempt to explain the negative
association between stock returns and inflation; and Fama and Gibbons (1982)
Evidence from an Emerging Stock Market
129
attempt to explain the negative association between inflation and real interest
rates.
Interest rates are among the most closely watched variables in the
economy. Their movements are reported almost daily by the news media.
They directly affect our everyday lives and have important consequences
for the health of the economy. The higher the interest rate, the higher
the discount factor, and lower the stock prices. Martikainen et al (1991)
used this variable in testing the APT model. The stock returns and
production growth, as outlined in Barro (1990) and Fama (1990), may be
affected by interest rates. Recently the boom in the Pakistani Stock Market
(KSE 100 index in all time highest in the recent past) is partly due to the
fact that interest rates in defense certificates and other interest-bearing
instrument have declined. Therefore investors are now coming to the
stock market, as a result demand for stock market securities is rising
which increase stock prices.
The regional market may have an effect on returns in the Karachi
Stock Exchange. In empirical studies many authors have used regional
market return as an independent variable, for example, for the case of the
Finland capital market (Helsinki Stock Exchange) Martikainen et al (1991)
have used the Stockholm Index.
Emerging Stock Market Factbook (1999) indicates that for the Pakistani
capital market the highest correlation of returns are with the Indian capital
market (0.40) and Malaysian capital market (0.36). Therefore in our analysis we
have used the Bombay Stock Index (BSE-30) and Kuala Lumpur Composite
Index (KLSE).
The U.S. stock market was by far the largest and most influential
capital market in the world. Therefore we have used Standard and Poor 500
index. S & P index reflects the worldwide expectations for all firms. The S &
P index has been selected since it is expected that the Pakistani stock
returns follow the global cash-flow expectations of firms. Rozeff (1984),
Shiller (1984) and Campbell and Shiller (1988) present evidence that
dividend yields forecast stock returns, Fama and French (1989) suggest that
dividend yields can explain cyclical variation in expected returns. Chen et al.
(1985) find that changes in aggregate production, inflation, and short-term
interest rates can explain the equilibrium pricing of equities, and Chen
(1991) shows that the cyclical behaviour of T-bill rates captures the cyclical
variation in equity risk premiums.
Javaid Iqbal and Aziz Haider
130
The money supply has typically been seen as a leading indicator, and it
is usually assumed that money supply and demand influence equity prices
(Fama, 1981; Geske and Roll, 1983; Kaul, 1987). The rise in money supply
can be expected to raise the stock prices (Martikainen et al., 1991). Kaul (1990)
also indicates significant association between monetary rule and the relationship
between stock returns and inflationary expectations. Monetary policy influences
stock returns by increasing future cash flows or by decreasing the discount
factors at which those cash flows are capitalized (Binswanger, 2000).
III. Exploratory Factor Analysis Approach
Our exploratory factor analysis approach is based on intuition, which
was presented by Chen et al. (1986) and which has been applied further by
several researchers. First, the factor scores and factor loadings from the
return series were estimated separately for the two subperiods and the
whole period. The estimation of factors was based on the principal
component method. Second, an orthogonal varimax rotation was applied. In
the following table, factors appear in decreasing order of variance explained
by the factors.
Table-1: Cumulative Proportions Of The Total Variance Explained By
Principal Components
Period
Fact 1 Fact 2 Fact 3 Fact 4 Fact 5 Fact 6 Fact 7 Fact 8 Fact 9
Jan-97--Dec-03 0.456 0.526 0.583 0.631 0.675 0.713 0.749 0.782 0.812
Jan-97--Dec-99 0.478 0.557 0.622 0.677 0.726 0.773 0.811 0.846 0.875
Jan-00--Dec-03 0.467 0.543 0.604 0.659
0.71
0.749 0.787 0.818 0.845
The cumulative proportions of the total variance explained by the
estimated factors are presented in Table 1. The results indicate that the figures
are quite stable over the two subperiods. We concentrate on nine factors
solution. This selection is based on the criteria of more than 80% variance
explained by the factors extracted. Using this criteria the Pre-specified
macroeconomic factors also support the existence of nine factors. The results of
the other estimated factor solutions are available from the authors on request.
Thus, the following nine-factor models were estimated for the stocks
to obtain asset sensitivities and unknown factors in the APT.
Rit − μ i = bi1 F1t + bi 2 F2t + .... + bi 9 F9t + eit
Evidence from an Emerging Stock Market
131
Where Rit , i = 1,2,...,24 , is the return of the stock i at month t, μ i
represent the mean return of the stock i, F1t , F2t ,..., F9t are the estimated
unknown common factors (factor scores), bi1 , bi 2 ,..., bi 9 are the asset
sensitivities (factor loadings) of the security i to the nine unknown factors,
and eit are the unsystematic return components of the stocks.
To test the linear risk-return relation implied by the APT, Table-2
presents OLS regressions where the estimated factor loadings are used as
independent variables, and the average returns of securities as dependent
variables.
Table-2: Regression Analysis Estimates For The Exploratory Factor
Analysis-Unrotated 9 Factors In The Model
Period Const Fact 1 Fact 2 Fact 3
Jan-97-- -0.007 0.0133 -0.0045
Dec-03 0.0008
T-value
-0.09 -0.55 1.45 -0.77
Jan-97-- 0.0197 0.02 -0.014 -0.007
Dec-99
T-value 1.67 1.27 -1.26 -0.7
Jan-00-- 0.0155 0.015 -0.002 -0.0093
Dec-03
T-value 1.91* 1.35 -0.28 -1.56
Fact 4 Fact 5 Fact 6 Fact 7 Fact 8 Fact 9 R2-adj
0.0030 0.0252 0.0112 -0.0095 -0.005 0.0197 41.20%
3
0.38 3.57*** 1.42 -1.31 -0.62 2.12*
0.0055 0.0109 0.0126 0.0094 -0.012 -0.006 16.10%
0.78 1.46 0.93 0.91 -1.06 -0.65
0.008 0.0124 -0.01 -0.007 -0.017 0.0053 42.00%
1.38
1.7
1.29
-0.98 -2.18** 0.65
Dependent variable: average monthly return for security; independent
variables: factor loadings.
* Significant at 0.10 level.
** Significant at 0.05 level.
*** Significant at 0.01 level.
The results indicate that in the whole sample period we can find two
priced factors according to this exploratory factor analysis approach; in the
first subperiod none of the factors seems to be priced, and in the second
subperiod we can find only one priced factor at the 5% significance level. The
number of priced factors seems to be very low and the results of this approach
indicate substantial instability of the explanatory power of the APT. This
instability may be due to a number of reasons as explained in section II.
IV. Pre-Specified Macroeconomic Factors Approach
Javaid Iqbal and Aziz Haider
132
Table-3 presents the principal components analysis on the 16 prespecified macroeconomic variables- the rotated solution. According to 80 % of
the variance explained criteria, the original variables were converted to 9
orthogonal time series. There are two reasons for the conversion. Firstly this
eliminates all problems with multi-colinearity and secondly it reduces the
dimensionality of the original variables and makes it easier to work with
time-series.
The factors in Table-3 appear in decreasing order of variance
explained by the factors, i.e. according to the eigenvalues of the factors. The
figures in the table are factor loading. Factor 1 indicates the real economic
activity, which are positively correlated. Factor 2 and factor 3 represented
the anticipated change and unanticipated change of inflation, which are also
positively correlated. Factor 4 represented stock index factor namely Bombay
Stock Exchange (BSE) and Karachi Stock Exchange (KSE) and dividend yield
with factor loadings 0.760, 0.509 and –0.834 respectively. Factor 5 and
factor 8 indicate clearly interest rate factor. Factor 6 indicates the stock
index factor Standard and Poor’s index (S&P) and BSE showing negative
correlation that is -0.906 and –0.487 respectively. Factor 7 represented the
money supply factor. Exchange rates are represented by factor 9 having
factor loading 0.976.
Tabel-3. Factor Pattern of the Macroeconomic Variables January-97 to
December-03
Variable
F1
F2
F3
F4
F5
F6
F7
F8
F9
Dmanu
0.89
-0.07
-0.08
-0.01
0.13
0.02
-0.20
0.04
-0.06
ddManu
0.87
0.03
-0.14
-0.10
0.00
-0.09 -0.11 -0.24
0.12
DKSE
-0.11
0.02
-0.08
0.76
0.07
-0.15 -0.03 -0.19
0.07
dS&P
0.03
-0.03
0.04
0.04
-0.01
-0.91
0.15
-0.07
DBSE
0.12
-0.17
-0.07
0.51
-0.06
-0.49 -0.40 -0.09
-0.02
DKLSE
0.02
-0.07
-0.15
0.20
-0.01
-0.03 -0.06
0.02
-0.06
dCPI-95
-0.02
0.17
0.91
-0.05
-0.15
0.03
0.10
-0.08
DdCPI
-0.19
0.11
0.90
-0.01
0.09
-0.06 -0.11 -0.01
0.01
dWPI-95
0.10
0.89
0.17
0.08
-0.07
0.07
0.05
0.05
-0.03
ddWPI
-0.15
0.91
0.09
0.06
-0.01
0.00
-0.10 -0.01
0.03
dExch.R
0.04
0.00
-0.06
0.01
-0.14
0.08
0.01
0.04
0.98
dT-bill
-0.01
-0.01
-0.02
0.01
-0.82
0.18
-0.14
0.30
0.10
dGovtB
-0.13
0.11
0.08
-0.08
-0.82
-0.22
0.13
-0.21
0.08
0.01
0.03
Evidence from an Emerging Stock Market
133
dMoneyM
-0.16
0.04
0.08
-0.19
-0.05
-0.15
0.89
0.04
dDividen
0.03
-0.20
-0.04
-0.83
-0.01
-0.12 -0.17 -0.01
0.07
dMoneyS
0.30
0.08
0.09
-0.09
-0.01
-0.02 -0.88 -0.01
-0.01
Variance
1.7746 1.7621 1.7501 1.6468 1.415 1.221 1.0746 1.0555 1.0156
Proportion 0.111
of Variance
DManu
0.02
0.11
0.109 0.103
0.088 0.076 0.067 0.066
0.063
= The change in the “real economic activity (manufacturing
index)”.
DdManu = The differentiated dmanu. This variable measures the
unanticipated change in the manufacturing index.
DKSE
= The change in the “Karachi Stock Exchange”.
dS&P
= The change in the “Standard and Poor 500 index”
dBSE
= The change in the Bombay Stock Exchange.
DKLSE
= The change in Kuala Lumpur Stock Exchange.
dCPI-95
= The change in the Consumer Price Index
ddCPI
= The differentiated dCPI-95. This variable measures the
unanticipated change in the Consumer Price Index.
dWPI-95 = The change in the Wholesale Price Index.
DdWPI
= The differentiated dWPI. This variable measures
unexpected change in the Wholesale Price Index.
dExch.R
= The change in the exchange rate between Pakistani rupee and
US dollar.
dT-bill
= The change in the Pakistani 90-day government treasury bills
return.
DGovtB
= The change in the Pakistani 10 years government bond return.
DMoneyM = The change in the money market rate.
DDividen = The change in the dividend yields.
DMoneyS = The change in the money supply.
the
134
Javaid Iqbal and Aziz Haider
The data have been obtained from various issues of International
Financial Statistics and Monthly Bulletin of State Bank of Pakistan. Stock
index data Bombay Stock Exchange and Karachi Stock Exchange data have
been obtained from www.scsecurities.com
To test the APT using pre-specified macro-economic factors, the
following time-series regressions were first estimated for the stocks to obtain
asset sensitivities and unknown factors in the APT.
Rit = α i + b1 F1 + b2 F2 + .... + bk Fk + eit
Where Rit is the return of the stock i at month t, α i is the intercept term
of the stock i, F j ( j = 1,2,..., k ) are in the above factor analysis estimated
macroeconomic factors (factor scores), bij ( j = 1,2,..., k ) are the sensitivities
of the return of the security i and eit are the unsystematic return
components of the stocks. In this OLS factor scores are used as independent
variables and stocks return for each stock as dependent variable. From this
we estimate factor sensitivity (factor loading).
Using these factor sensitivities as independent variable and stock
average returns as dependent variable, the following regression was run
Ri = Lo + L1b1 + L2 b2 + ... + LK bK
This estimated risk premium L’s and tested which factors were
priced. The results of this regression are reported in Table-4.
Table-4: Regression Analysis Estimates For The Pre-Specified Factors
Approach:- Rotated
Period Const Fact 1 Fact 2 Fact 3 Fact 4 Fact 5 Fact 6 Fact 7 Fact 8 Fact 9 R2-adj
Jan-97– 0.0171 0.0423 -0.0725 0.4283 -0.2313 0.1842 0.031 0.0589 0.0650 0.2164 39.30
Dec-03
6
%
T-value
3.8*** 0.31 -0.574 2.38** -3.7*** 0.96 0.24 0.48 0.48 1.61
Jan-97– 0.0177 0.0453 -0.1689 0.1366 0.2598 -0.0186 0.030 0.0441 -0.0095 -0.0152 48.20
Dec-99
6
%
T-value
5.04*** 0.4600 -1.1100 3.9*** 2.3300* -0.1500 0.230 0.4500 -0.1000 -0.1800
*
0
Evidence from an Emerging Stock Market
135
Jan-00– 0.0134 -0.1575 0.1429
-0.2275 0.1082 0.059 0.1196 0.0389 0.0782 7.40%
Dec-03
0.0358
0
T-value
1.7000 -1.4700 0.8300
-1.0000 0.7800 0.220 0.9600 0.3100 0.3700
0.1900
0
Dependent variable:- average monthly return for security.
Independent variables:- sensitivities of asset returns to changes in macroeconomic
factors.
* Significant at 0.10 levels.
** Significant at 0.05 levels.
*** Significant at 0.01 levels.
The results imply that we can find two priced factor in the whole
sample period when factor 3 and 4 become priced. The first sample period
also shows the same result but this time the only change is the significance
level is reversed. We know from the analysis in the preceding step that the
third factor is the anticipated and unanticipated inflation and the fourth
factor is the stock market index and dividend yield. The second sample
period shows no priced factor. As value of the intercept is significant, it is
likely that there are other factors which are not included in our study, that
affect stock returns and also are priced. Again, quite a high level of
instability is found in the results.
V. Conclusions
The results of two different testing methods for the Arbitrage
Pricing Theory (APT) are nearly the same because in the whole sample
period two priced factors are found. This is an encouraging result, which
supports the theory. But the number of priced factors seems to be very low
and the results of both approaches indicate substantial instability of the
explanatory power of the APT. Explanatory factor analysis approach indicates
two factors governing stock return. Pre-specified macro economic approach
identifies these two factors as the unanticipated and anticipated inflation,
market index and dividend yield. The former factor was also identified by
Attaullah (2001). Some evidence of instability is found. In the second
subperiod namely January 2000 to December 2003 that is more volatile,
the APT based on exploratory factor analysis on stock returns performs
relatively well. In the first subperiod extending from January 1997 to
December 1999 the APT based on pre-specified macroeconomic variables is
supported.
136
Javaid Iqbal and Aziz Haider
References
Abeysekera, S. P. and Mahajan, A. 1988, A test of the APT using UK stocks,
Journal of Business Finance and Accounting, 14, 377-391.
Ahmad, E. and Rosser, B.J. 1995, Non-linear speculative bubbles in the
Pakistan
stock markets Pakistan Development Review, 34 (1),
25-41.
Ahmad, E. and Zaman, B.U. 2000, Risk, uncertainty and return at Karachi
Stock Exchange, The Lahore Journal of Economics, 5(2) 107-126.
Asprm, M. 1989, Stock prices, asset portfolios and macroeconomic variables
in ten European countries, Journal of Banking and Finance, 13,
589-612.
Attaullah, A. 2001, Macroeconomic variables as common pervasive risk
factors and empirical content of the Arbitrage Pricing Theory in
Pakistan, The Lahore Journal of Economics, 6 (1) 55-77.
Barro, R. 1990, The stock market and investment, Review of Financial
Studies, 3, 115-31.
Binswanger, M. 2000, Stock returns and real activity: is there still a
connection?, Applied Financial Economics, 10, 379-387.
Booth et al. 1993, On the Existence of Common Factors in the Arbitrage
Pricing Model: Internaltional Evidence From US and Scandinavian
Stock Markets, Applied Financial Economics, 3.
Campbell, J.Y. and Shiller, R. 1988, The dividend-price ratio and
expectations of future dividends and discount factors, Review of
Financial Studies, 1, 195-228.
Chen, N. 1991, Financial investment opportunities and macroeconomy,
Journal of Finance, 46, 529-54.
Chen, N., Chan, K. and Hsieh, D. 1985, An exploratory investigation of the
firm size effect, Journal of Business, 56, 383-403.
Chen, N.F. 1983, Some Empirical Tests of the Theory of Arbitrage Pricing,
Journal of Finance, Vol XXXVIII, No 5.
Evidence from an Emerging Stock Market
137
Chen, N-F. 1989, Financial investment opportunities and the real
economy. Paper presented to the 16th Meeting of the European
Finance Association.
Chen, N-F., Roll, R. and Ross, S.A. 1986, Economic forces and the stock
market, Journal of Business, 59, 383-403.
Copeland, T. E. and Weston, J. F. 1988, Financial theory and corporate
policy, Addison-Wesley, Reading, Ma.
Diacogiannis, G. P. 1986, Arbitrage Pricing Model: a critical examination
of its empirical applicability for the London Stock Exchange,
Journal of Business Finance and Accounting, 12, 489-504.
Emerging Stock Market Factbook
Corporation, Washington.
1999,
International
Finance
Fama, E. F. 1981, Stock returns, real activity, inflation, and money,
American Economic Review, 71, 545-65.
Fama, E. F. 1990, Stock returns, expected returns, and real activity,
Journal of Finance, 45, 1089-180.
Fama, E. F. and French, K. R. 1989, Business conditions and expected
returns on stocks and bonds, Journal of Financial Economics, 25,
23-49.
Fama, E. F. and Michad, R. Gibbons 1982, Infaltion , real returns and
capital investment, Journal of Monetary Economics, 9, 297-323
Geske, R. and Roll, R. 1983, The fiscal and monetary linkage between
stock returns and inflation, Journal of Finance, 38, 1-38.
Hussain, F. 2000, The day of the week effect in the Pakistan equity
market: An investigation, The Lahore Journal of Economics, 5(1),
93-98
Hussain, F. and Uppal, J. 1998, Distribution of stock returns in an
emerging market: The Pakistani market, Pakistan Economic and
Social Review,36 (1), 47-72.
Kaul, G. 1987, Stock returns and inflation. The role of the monetary
sector, Journal of Financial Economics, 18, 253-276.
138
Javaid Iqbal and Aziz Haider
Kaul, G. 1990, Monetary regimes and the relation between stock returns
and inflationary expectations, Journal of Financial and Quantitative
Analysis, 25, 307-321.
Khilji, N. M. 1993, The behaviour of stock return in an emerging market: A
case study of Pakistan. Pakistan Development Review, 32 (4), 593-604.
Khilji, N. M. 1994, Nonlinear dynamics and chaos: Application to financial
market in Pakistan. Pakistan Development Review, 33 (4), 14171429.
Lintner, J. 1965, The Valuation of Risk Assets and the Selection of Risky
Investments in Stock Portfolios and Capital Budgets, Rev. Econ.
Statistics, 47(1).
Loflund 1992, Arbitrage Pricing Theory in a Small Open Economy-Empirical
Evidence from the Swedish Stock Market, Forskningsrapport,
Svenska Handelshogskolan, Helsingfors.
Martikainen, T., Yli-Olli, P. and Gunasekaran, A. 1991, Incremental
significance of pre-specified macroeconomic factors in testing the
arbitrage pricing theory: empirical evidence with Finnish data,
Applied Financial Economics, 1, 139-147.
Mishkin, F. S. 1997, The Economics of Money, Banking, and Financial
Market, Addison Wesley Longman Inc.
Mossin, J. 1966, Equilibrium in a Capital Asset Market, Econometrica, 34,5.
Ostermark. 1989, Arbitrage Pricing Models for Two Scandinavian Stock
Markets, Omega International Journal of Management Science.
Ram, R. and David E. S. 1983, Stock returns, real activity, inflation, and
money: comment, American Economic Review, 73, 463-470.
Roll, R. and Ross, S. A. 1980, An empirical investigation of the Arbitrage
Pricing Theory, Journal of Finance, 35, 1073-1103.
Ross, S. A. 1976, The Arbitrage Pricing Theory of capital asset pricing,
Journal of Economic Theory, 13, 341-360.
Rozeff, M. 1984, Dividend yields are equity risk premiums, Journal of
Portfolio Management, 68-75.
Evidence from an Emerging Stock Market
139
Rubio, G. 1988, Further international evidence on asset pricing: the case of
Spanish capital market, Journal of Banking and Finance, 12, 221-242.
Sharpe, W. F. 1964, Capital Asset Prices: A theory of market equilibrium
under conditions of risk, Journal of Finance 19(5).
Shiller, R. J. 1984, Stock prices and social dynamics, Brookings Papers on
Economic Activity, 2, 457-510.
Stulz,R. M. 1986, Asset pricing and expected inflation, Journal of Finance,
41, 209-223.
Wasserfallen, W. 1989, Macroeconomic news and the stock market. Evidence
from Europe, Journal of Banking and Finance, 13, 613-626.
Yli-Olli, P. and Virtanen, I. 1989, Arbitrage Pricing Theory and its empirical
applicability for the Helsinki Stock Exchange, European Institute for
Advanced Studies in Management, Working Paper 89-07.
Yli-Olli, P., Virtanen I. and Martikainen, T. 1990, Common factors in the
Arbitrage Pricing Model in two Scandinavian countries, OMEGA
International Journal of Management Science, 18, 615-624.